flowchart TD
subgraph 01_load
%%load
R{{"download_dataset_ncbi"}} -->|"read_unstructured_ncbi_table"| A[ncbi_data]
T{{"download_data_annotation_ncbi"}} -->|"read_bgx_file"| I[annotation_data]
%%write
A[ncbi_data] -->|"write_tsv"| B[01_ncbi_data.tsv.gz]
I["annotation_data$probes"] --> |"write_tsv"| J["01_ncbi_annot_probes.tsv.gz"]
end
subgraph 02_clean
%%ncbi_data pipe
B[01_ncbi_data.tsv.gz] -->|"read_tsv"| C[ncbi_data]
C[ncbi_data] -->|"as_tibble |> <br/>filter(!str_detect) |> <br/>slice(41:n) |> <br/>t()"| D[ncbi_dirty_pheno]
D[ncbi_dirty_pheno] --> |"colnames[1, ] |> <br/>str_replace_all('!sample_', ' ')"| E[ncbi_dirty_pheno]
E[ncbi_dirty_pheno] --> |"[-1, ] |> <br/>as_tibble() |> <br/>rename_with()"| F[ncbi_dirty_pheno]
F[ncbi_dirty_pheno] --> |"select() |> <br/>mutate() |> <br/>select() |> <br/>t() |> <br/>as.data.frame() |> <br/>rename_with() |> <br/>as_tibble()"| H[ncbi_clean_pheno]:::red
%%sample source and all of the expr
H[ncbi_clean_pheno]:::red --> |"select() |> <br/>mutate() |> <br/>rename() |> <br/>select()"| S[sample_source table]
S[sample_source table] --- U>"get_gene_expr_by_source()"]
Q["ncbi_clean_expression"] --- U>"get_gene_expr_by_source()"]
U>"get_gene_expr_by_source()"] --> V[gene_expr_placenta]:::red
U>"get_gene_expr_by_source()"] --> W[gene_expr_maternal_blood]:::red
U>"get_gene_expr_by_source()"] --> X[gene_expr_cord_blood]:::red
%%ncbi_dirty_expr pipe
C[ncbi_data] -->|"slice()"| G[ncbi_dirty_expr]
G[ncbi_dirty_expr] --> |"colnames() |> <br/>slice() |> <br/>unlist()"| L[ncbi_dirty_expr]
L[ncbi_dirty_expr] --> |"rename() |> <br/>mutate(across(-probe.id, ~ as numeric(.x))) "| M[ncbi_dirty_expr]
%%annotation probe
J["01_ncbi_annot_probes.tsv.gz"] -->|"read_tsv()"| N["ncbi_annot_probes"]
N["ncbi_annot_probes"] --> |"select()"| O["gene_annotation"]
%%merging to clean ncbi
O["gene_annotation"] ---|"select()"| P( ):::empty
M["ncbi_dirty_expr"] --- P( )
P( ) -->|"right_join(by = c('Probe_Id' = 'probe_id')) |> <br/>rename() |> <br/>select()"| Q["ncbi_clean_expression"]:::red
classDef red stroke:#f00,stroke-width:3px
end
%%writing and reading
%%subgraph 04_augment
%%Q["ncbi_clean_expression"] --> Y[gene_expr_placenta]
%%V[gene_expr_placenta] --> Z[gene_expr_placenta]
%%W[gene_expr_maternal_blood] --> AA[gene_expr_maternal_blood]
%%X[gene_expr_cord_blood] --> AB[gene_expr_cord_blood]
%%H[ncbi_clean_pheno] --> AC[ncbi_clean_pheno]
%%end
